[mythtv-users] Help with backtrace - Does this look right?

[Tom Lichti]

On Wed, May 2, 2012 at 10:19 AM, A McDermott <list at mcdermotts.ca> wrote:
> I am trying to capture a backtrace of a segfault within libQtcore.so.4
> on my backend.  For example:
>
> May  1 13:59:54 pvr kernel: [173048.994311] mythbackend[13043]:
> segfault at 48 ip 00007f1e0d4d8cbd sp 00007f1dfdbc4cd0 error 4 in
> libQtCore.so.4.7.4[7f1e0d359000+292000]
>
> When logs rotated today, gdb caught the HUP and printed a backtrace -
> which I understand.  What I am curious about is the number of "No
> Symbol table info available" entries in the backtrace entries - see
> "Thread #3" below.  Some threads do have symbols available (looks like
> from within myth itself not the libraries - see "thread #1" below)
>
> My question is does this look correct?  When gdb finally captures the
> actual segfault, will this be sufficient information for the ticket?
> Or, do I need to do something else to get symbols for the various
> libraries?
>
> Thread 3 (Thread 0x7fffdffff700 (LWP 1817)):
> #0  0x00007ffff5fff3cb in pthread_cond_timedwait@@GLIBC_2.3.2 () from
> /lib/x86_64-linux-gnu/libpthread.so.0
> No symbol table info available.

I believe that for any system libraries (like libpthread) you need to
install the -debug version of the library. I could be wrong though.

Tom